0
\$\begingroup\$

I have an stm32f407 which was running okay for like a month. but now i am seeing that whenever I connect power supply the current consumption is very high and the MCU is heating up a lot. The decoupling capacitors on the Vdd pins show short to ground with multimeter. I removed 3 of them and checked them with multimeter but they were okay so I placed them back now. Should I go on and remove the rest of the decoupling capacitors and test one by one or the problem is coming from somewhere else? I have a power supply which provides regulated 3.3V and is very reliable.

Also for information, this PCB has two sources of power inputs, one is from 12 volt power supply and the other one is from 5V USB. I have two buck converters on this custom PCB and I recently found out that the 5V buck converter has no reverse polarity protection and because of that whenever I connected the USB power and didn't connect the power supply there was this chattering noise coming from the PCB, what I did was that I separated the 5V USB from the 5V buck's output with a knife. And then the high frequency noise disappeared. To remove the possibility that my power circuit may be bad I have disconnected both the buck converters. And connected a 3.3V powersupply directly on the 3.3V trace. So right now I have only two sensors connected to the MCU one is ICM42688 and another one is MS5611. And both works as required and not heating up (I can't feel it with the finger test).

So now I think I have narrowed down the problem to the MCU and the decoupling capacitors, either the MCU is bad or one of the capacitor has gone bad because of that reverse polarity on the buck converter.

Is there anyone who has any experience and can guide me? should I remove the other capacitors as well and check one by one or is there anything else that can be done?

Schematics and PCB layout: enter image description here

I removed the traces in pink and placed 3.3V power on the point in green.

enter image description here

enter image description here

enter image description here

this is how the VBUS and the PA9 pins are connected. the VBUS goes nowhere else than lighting up the LED and the PA9 pin. And I have checked just now the PA9 pin is not working. but I still am confused that why it got damaged.

\$\endgroup\$
17
  • 5
    \$\begingroup\$ If it's the MCU that's heating up then it's 99.99% certain that it's the MCU that's bad (not the decoupling caps). \$\endgroup\$
    – brhans
    Sep 19, 2023 at 16:07
  • 6
    \$\begingroup\$ It may be that a lot of current is passing through the MCU. It may be shorted inside it or some external device on IO pins pushes a lot of current into MCU or draws a lot of current out of MCU. \$\endgroup\$
    – Justme
    Sep 19, 2023 at 16:15
  • 1
    \$\begingroup\$ how would the decoupling caps draw current through the MCU? ... how do you know that the MCU is working properly? \$\endgroup\$
    – jsotola
    Sep 19, 2023 at 16:42
  • 1
    \$\begingroup\$ @MirHamza No, it's not possible that a bad decoupling cap would cause a lot of current to flow from the VDD pins of the uC, causing it to heat up. A bad cap could draw current from the supply, causing the cap to heat up. \$\endgroup\$
    – John D
    Sep 19, 2023 at 16:43
  • 1
    \$\begingroup\$ The new information that PA9 isn't working does seem to confirm that the failure came from connecting 5V at VBUS with no power at VDD. The circuitry for that pin is damaged and probably the source of heat. If you remove 5V from PA9 the rest of the chip may keep working; the longer you let it overheat the higher chances that the damage will spread. \$\endgroup\$
    – Ben Voigt
    Sep 19, 2023 at 17:24

1 Answer 1

0
\$\begingroup\$

The problem is you have PA9 connected to VBUS. VBUS is usually 5V, that means the protection diodes will be turning on and sinking a lot of current from VBUS. You need a resistor at minimum. The ports on a STM32 are 5V tolerant, but that doesn't mean they won't sink a lot of current from a power supply rail.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.